On Fri, Dec 23, 2011 at 12:19:53AM -0800, Doug Barton wrote:
> Wrong answer. The right answer is for DHCPv6 to be a superset of RA so
> that organizations that need more than RA provides only have to run
> DHCP. As a free bonus, it's FAR easier to add one knob (default route)
> to DHCPv6 than it is to add all features to RA. Extra bonus, the only
> hurdles are political, not technical.

I agree with your essential point, except to say that for me to be happy
with DHCPv6, I also want to be able to assign an address to a host by
taking some constant bit of information from the host (like, say, a MAC
address) and know that no matter how I reinstall that host, it'll retain
the address I gave it.

As it is today, to use DHCPv6 the way I want, I must first have my host
get an IP (the wrong one), figure out what DUID it used, and reconfigure
my server to give it the one I want it to have. With RA, I don't assign
the address, but to know what it is, I just look up its MAC address and
I'm done. With DHCPv4, I look up its MAC address, add it to my server
configuration and I'm done.

I don't really care if it's a MAC address per se, but I don't want to
have to deal with DUID change because of OS reinstallation, change of
client software, dual-boot, removal & re-adding of the connection
definition, etc. I can think of reasons why it might be nice to have
different DUIDs for different OS's, so I don't object to the option, but
in most cases the current spec hurts (me) more than it helps.

-- 
Scott Schmit

Attachment: smime.p7s
Description: S/MIME cryptographic signature

--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to